The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to execute ffmpeg and ffprobe for audio processing and video rendering in scripts/generate_voiceover.py and SKILL.md.
[COMMAND_EXECUTION]: It invokes npm and npx commands (e.g., create-video, remotion studio) to initialize and manage the video project.
[EXTERNAL_DOWNLOADS]: Fetches synthesized audio from the ElevenLabs API using urllib.request.urlopen in scripts/generate_voiceover.py.
[EXTERNAL_DOWNLOADS]: Recommends downloading royalty-free background music from well-known services like Pixabay and Bensound via curl as part of the setup workflow.
[EXTERNAL_DOWNLOADS]: Automates the installation of necessary dependencies such as openai-whisper via pip and lucide-react via npm.
[DATA_EXFILTRATION]: Transmits the user-defined voiceover script and the user's ElevenLabs API key to api.elevenlabs.io for text-to-speech generation, which is the primary intended functionality.
[PROMPT_INJECTION]: The skill analyzes local project metadata (e.g., git log, README.md) to generate video content suggestions, creating a surface for indirect prompt injection from malicious files within a repository.
Ingestion points: The skill reads the output of git log --oneline -100 and the first 30 lines of README.md to define product context.
Boundary markers: No explicit delimiters or instructions are used to separate ingested file content from agent instructions.
Capability inventory: Access to Bash (including ffmpeg, pip, npm), file system Read/Write, and network operations.
Sanitization: No sanitization of the analyzed repository content is performed before it is processed by the agent.

promo-video